Published data for four cyclone designs were analysed with multi-linear regression, in order to determine which parameters influenced the cyclone collection efficiency. The dependence of the cyclone cut-off size on fluid how was expressed by two parameters (the ratio of the cut-off size to the cyclone body diameter, and a Reynolds number based on the inlet velocity and the annular distance between the vortex tube and the cyclone wall) originally proposed by Moore and McFarland. The steepness of the collection efficiency curve was found to be mainly influenced by the ratio of the cyclone body diameter to the vortex tube inner diameter, and independent of the Reynolds number. A literature review provided an explanation of the different behaviour between long- and short-coned cyclones,which stems from differences in the vortex flow: At low Reynolds numbers there is practically no turbulence, at intermediate Reynolds numbers the Vortex is distorted by periodic waves and is attached to the cyclone wall, and at high Reynolds numbers the turbulent mixing is considerable and the vortex is attached to the cyclone bottom. The transition between the intermediate and the high-Reynolds number vortex flow regimes is determined by the Reynolds number based on the cyclone inlet.